Sprint ##DATA tuning (changing Home Agent)

scott74 · EVDO Fledgling Joined: 09 Apr 2006 Posts: 22 Location: STL

This is great, I tried a lot of the addresses posted here and their where dramatic differences. Finally a solution, I was only averageing 250k a lot of times, called tech support to se if their was anything they could do on their end and they just said no. I am trying to stream over a slingbox and sometimes the slow speeds just were not cutting it.

HA's that were assigned to me 68.28.81.76 and 68.28.89.76 speeds seem to be in the 200's with these, sometimes in the 500's, this is in the St Louis are.

Last night switched to 68.28.49.76 and 68.28.89.77 and speeds started going crazy rangeing from 600k to 1.2k Don't know if this was a fluke or what but I liked the results. This was in the Fenton MO area. 5 to 6 bar signal. Got to the Arnold Mo area (18 miles from Fenton) and speeds seemed to hold at 500k inside house. The fenton test where outside. I'll keep plalying with them.

Great post Xenophon

xenophon · EVDO Addict Joined: 30 Aug 2005 Posts: 1969

That's odd that the KC and Omaha HA nodes (consistently the fastest) were slower for you from St. Louis and that the W Coast was faster... but whatever works for you, cool.

I'm looking for more HA subnets. May as well collect the whole set of Pokemon cards.

pverzoni · EVDO Newbie Joined: 30 Mar 2006 Posts: 10

If I change the HA address, then move the card to a KR1, will the HA stick or will it only work when using the Sprint Connection Manager?

xenophon · EVDO Addict Joined: 30 Aug 2005 Posts: 1969

Yes, it sticks with the card and not Connection Manager. To confirm, you can do traceroutes when in the KR1 and determine the HA from the traceroute posting described on page two of this thread.

But the HA will reset to default on your account if you upgrade the firmware on the card, reactivate it or do an Update Profile within Connection Manager.

Sprint seems to have improved all HA nodes but it's still possible some will perform better than others depending on your location. And if the billing address is distant from your most common location, the HA may not be in your region.

BTW, I scripted to ping all subnets of the 68.28.xx.76 network and found no other HA's. The 8 and 30 subnets were pingable but apparently they are not HAs as I would get a login failed if attempting to use them. So methinks that all HAs listed in the first post of this thread are all of them.

pverzoni · EVDO Newbie Joined: 30 Mar 2006 Posts: 10

This works great!

I'm in NYC and lately have been well below 500 kb/s - sometimes during the day I would barely get above 200 kb/s.

So I logged in as per your suggestion and saw my primary was Pensylvania and secondary something else (not NYC). I changed primary to NYC and I am getting EVDO again (around 700kb/s).

Now what happens if I travel to LA, how do the packets flow? Do they go from the tower to NYC then back to the LA tower (does this make sense)? Just curious..

I am so glad I tried this out!

xenophon · EVDO Addict Joined: 30 Aug 2005 Posts: 1969

Very cool that it makes a difference for you!

Unfortunately when you travel, the HA address does not change to one near where you travel. If you are in LA and have a NYC HA, the traffic likely significantly bounces around, so you may want to switch HA to the region you are visiting if performance is not acceptable.

scott74 · EVDO Fledgling Joined: 09 Apr 2006 Posts: 22 Location: STL

this sucks, i was getting good results and know i can't get over 100k while getting a 6 bar signal and a -71db consistantly. does anybody know whats going on?

IdioTunes · EVDO Fledgling Joined: 19 Apr 2006 Posts: 17

you are on 1xRTT.

anyhow, im in a bad signal area ATM (-100dBM) and am pulling down 600-700kbps Smile

xenophon · EVDO Addict Joined: 30 Aug 2005 Posts: 1969

Sitting at a conference in Vegas and testing from a conf room with different home agents.

Take a look at results here...

http://www.dslreports.com/archive/spcsdns.net?zip=89119&start=Search

The distance from HA does not seem to impact performance, it appears to be more load at the time of day. From Vegas, a New York HA can perform as well as California HA depending on load and time of day, over 1Mbps. (I used a test site nearest to each HA location).

But there is a clear impact on ping times. From Vegas, if I ping web sites on the W coast using western half HAs, the ping is typically 150-200ms. If I ping web sites on E coast using E coast HAs, the ping is typically 250-300ms. If I ping web sites on W coast using E coast HAs, the ping time is usually closer to 300ms.

So the HA can perform well even on other coast depending on time of day, but not surprisingly, there is a correlation to distance from HA and latency.

scott74 · EVDO Fledgling Joined: 09 Apr 2006 Posts: 22 Location: STL

Is their a chance that you can pick up 1Xrtt even though I have my card locked to EVDO. Last night at work when I was in the front of the building I was getting between 800k and 1000k, when I walk approx. 500 feet to the back of the building I was only pulling around 150k, same as the middle of the building I get 350k and down, most of the time. Have not seen the high speeds in the middle of building.

Sprint said that I am a mile within EVDO coverage. Could it be that 200 to 500 feet could be the difference between EVDO and 1Xrtt.

P.S. I was pulling the faster speeds with less of a signal. When I am inside the building I am using a * amp and magnet antenna on the roof, (6bars and -71db). This is on K.C. HA's from St Louis area.

IdioTunes · EVDO Fledgling Joined: 19 Apr 2006 Posts: 17

http://www.dslreports.com/speedtests/1503671;80590;dd6ec1477339d2d1d15b1ada35f8fe85;3.0;www.dslreports.com/1146769007

br0adband · Posted: Sat May 06, 2006 4:46 pm Post subject:

New member, first posting.

I'm in downtown Las Vegas, about 2 blocks from the Sprint building at the corner of Fremont and Las Vegas Blvd. I can see the cellsite tower on the building from my balcony, actually, so my assumption is that is what I'm connecting to using this Sprint PC-5740 I just got on Thursday evening.

The service works great so far, and I'm pretty satisfied with it. Average download speeds are showing in the 700-800Kbps range as expected, uploads are anywhere from 70 to 128-ish.

I've spent the past hour or so reading postings here and I'm interested in trying this HA tuning myself but I can't seem to narrow down any specific postings about changing the info. I've tried hovering the mouse cursor over the Go button on Connection Manager when I'm disconnected, but I never get any data entry popups, nor does pressing the ##code work, and I've tried it many times.

Without sounding like a total n00b here, can someone provide a simple answer as to how I go about doing the tuning itself and perhaps provide a list of the HA's that I can use? I'm gunning for that mythical >1Mbps speed just to see if I can do it.

Not sure where to measure the bars of signal strength, but in the Device Info & Diagnostics information I've got a solid -73dBm. Funny thing though, when I fired it up earlier today when I woke up, the signal strength was showing -62dBm for a while, then it jumped back up to -73dBm where it currently sits.

I've taken the laptop out on the balcony so there is nothing in between the card's antennas and the cellsite itself, direct line-of-sight, and curiously I got no change in speeds or signal strength. That makes me wonder if that site is the one I'm connecting to; I make the assumption that it is since it's the closest one and I'm damned near on top of it.

Thanks for any assistance and thanks for providing this resource for EVDO wireless users.

Paul

EDIT:

Duh, didn't realize the letters of the code had to be in CAPS. Smile

Just tried to do the ##DATA (in Caps of course) and there the popup is, silly me. Now to figure out which HA's to try.

br0adband · Posted: Sat May 06, 2006 5:17 pm Post subject: The King of all Speedtests

Not sure if it's here at the EVDO forums, but I see a lot of postings pointing to DSL Reports, Frontier.net and the Speakeasy speedtests, and those are just fine for quick down and dirty graphical speedtest results.

But if you're interested in the real nuts and bolts and more statistical data than you can throw a stick at, try this one:

http://ciseweb100.cise-nsf.gov:7123/

There are links to other similar speedtests at the bottom of the page (in case you want to do it by geographical location to find something closer to your spot), but as far as being the biggest baddest speedtest on the entire Internet, these pretty much cover it.

Hope this helps,
Paul

digitrip · EVDO User Joined: 04 May 2006 Posts: 35

That link didn't work for me. Can you check it again? (I tried taking the port off too, but that didn't work either)

br0adband · Posted: Fri May 12, 2006 12:17 am Post subject:

Try this one:

http://jlab4.jlab.org:7123/

They're all cross-linked to each other at the bottom of the pages. Not sure why that primary one isn't working, could be down for maintenance or whatever. The Jefferson Labs one works, because I just tried it.

Just to give everyone an idea of what this speedtest provides, here are the 3 primary outputs you get:

First is the main screen info and the test I just ran looks like this:

TCP/Web100 Network Diagnostic Tool v5.2.0f
click START to begin
Checking for Middleboxes . . . . . . . . . . . . . . . . . . Done
running 10s outbound test (client to server) . . . . . 122.75Kb/s
running 10s inbound test (server to client) . . . . . . 1.11Mb/s
Your PC is connected to a Cable/DSL modem
Information: Other network traffic is congesting the link

Second is the Statistics window accessed by clicking that button:

WEB100 Enabled Statistics:
Checking for Middleboxes . . . . . . . . . . . . . . . . . . Done
running 10s outbound test (client to server) . . . . . 129.32Kb/s
running 10s inbound test (server to client) . . . . . . 1.11Mb/s

------ Client System Details ------
OS data: Name = Windows XP, Architecture = x86, Version = 5.1
Java data: Vendor = Sun Microsystems Inc., Version = 1.5.0_06

------ Web100 Detailed Analysis ------
Cable modem/DSL/T1 link found.
Link set to Half Duplex mode
Information: throughput is limited by other network traffic.
Good network cable(s) found
Normal duplex operation found.

Web100 reports the Round trip time = 290.23 msec; the Packet size = 1380 Bytes; and
There were 7 packets retransmitted, 37 duplicate acks received, and 41 SACK blocks received
The connection was idle 0 seconds (0%) of the time
This connection is network limited 99.68% of the time.
Contact your local network administrator to report a network problem

Web100 reports TCP negotiated the optional Performance Settings to:
RFC 2018 Selective Acknowledgment: ON
RFC 896 Nagle Algorithm: ON
RFC 3168 Explicit Congestion Notification: OFF
RFC 1323 Time Stamping: OFF
RFC 1323 Window Scaling: ON
Information: Network Middlebox is modifying MSS variable
Server IP addresses are preserved End-to-End
Client IP addresses are preserved End-to-End

Third is the "More Details" by clicking that button:

WEB100 Kernel Variables:
Client: localhost/127.0.0.1
AckPktsIn: 459
AckPktsOut: 0
BytesRetrans: 9660
CongestionSignals: 1
CountRTT: 418
CurrentCwnd: 48300
CurrentMSS: 1380
CurrentRTO: 510
CurrentRwinRcvd: 256960
CurrentRwinSent: 5840
CurrentSsthresh: 30360
DSACKDups: 0
DataBytesIn: 0
DataBytesOut: 1257180
DataPktsIn: 0
DataPktsOut: 911
DupAcksIn: 37
ECNEnabled: 0
FastRetran: 1
MaxCwnd: 62100
MaxMSS: 1380
MaxRTO: 890
MaxRTT: 510
MaxRwinRcvd: 256960
MaxRwinSent: 5840
MaxSsthresh: 75900
MinMSS: 1380
MinRTO: 470
MinRTT: 210
MinRwinRcvd: 64240
MinRwinSent: 5840
NagleEnabled: 1
PktsIn: 459
PktsOut: 911
PktsRetrans: 7
Rcvbuf: 128000
SACKEnabled: 3
SACKsRcvd: 41
SmoothedRTT: 280
Sndbuf: 128000
SndLimTimeRwin: 0
SndLimTimeCwnd: 10076286
SndLimTimeSender: 31848
SndLimTransRwin: 0
SndLimTransCwnd: 1
SndLimTransSender: 1
SndLimBytesRwin: 0
SndLimBytesCwnd: 1257180
SndLimBytesSender: 0
SumRTT: 121320
Timeouts: 0
TimestampsEnabled: 0
WinScaleRcvd: 2
WinScaleSent: 7
DupAcksOut: 0
StartTime: 2147483647
CurrTime: 10113367
c2sData: 2
c2sAck: 2
s2cData: 9
s2cAck: 2

Checking for mismatch condition
(cwndtime > .3) [0.99>.3], (MaxSsthresh > 0) [75900>0],
(PktsRetrans/sec > 2) [0.69>2], (estimate > 2) [0.44>2]
Checking for mismatch on uplink
(speed > 50 [0.99>50], (xmitspeed < 5) [0.12<5]
(rwintime > .9) [0>.9], (loss < .01) [0.00<.01]
Checking for excessive errors condition
(loss/sec > .15) [6.58>.15], (cwndtime > .6) [0.99>.6],
(loss < .01) [0.00<.01], (MaxSsthresh > 0) [75900>0]
Checking for 10 Mbps link
(speed < 9.5) [0.99<9.5], (speed > 3.0) [0.99>3.0]
(xmitspeed < 9.5) [0.12<9.5] (loss < .01) [0.00<.01], (link > 0) [0>0]
Checking for Wireless link
(sendtime = 0) [0.00=0], (speed < 5) [0.99<5]
(Estimate > 50 [0.44>50], (Rwintime > 90) [0>.90]
(RwinTrans/CwndTrans = 1) [0/1=1], (link > 0) [0>0]
Checking for DSL/Cable Modem link
(speed < 2) [0.99<2], (SndLimTransSender = 0) [1=0]
(SendTime = 0) [0.003150729897328231=0], (link > 0) [0>0]
Checking for half-duplex condition
(rwintime > .95) [0>.95], (RwinTrans/sec > 30) [0>30],
(SenderTrans/sec > 30) [0.10>30], OR (link <= 10) [0<=10]
Checking for congestion
(cwndtime > .02) [0.99>.02], (mismatch = 0) [0=0]
(MaxSsthresh > 0) [75900>0]

bw = 0.44 based on packet size = 10Kbits, RTT = 290.23msec, and loss = 0.006586168892681599
The theoretical network limit is 0.44 Mbps
The transmit buffer (125.0 KByte) limits the application to 3.36 Mbps
Your receive buffer (250.0 KByte) limits the application to 6.75 Mbps
The network based flow control limits the application to 1.63 Mbps

Client Data reports link is 'T1', Client Acks report link is 'T1'
Server Data reports link is '10 Gig', Server Acks report link is 'T1'

So, as I stated above, this speedtest gives you far more information than any other speedtest I've ever found on the Internet, and all of it is actually useful and relatively easy to understand.

By the way, in the week I've owned this Sprint EVDO card, that's the first time I've tested at over a megabit per second... must be a good night! Smile

Hope this helps...
bb

ps
One other interesting statistic: that Jefferson Labs server is located in Newport News, VA. I've actually been on the campus of the labs in the past (I'm from Portsmouth, VA, originally). Currently I'm in Las Vegas, NV, soooo... that's a pretty damned good speed considering the distance covered I'd say.